Marketing attribution is how you assign credit for a sale or a lead to the marketing touches that led to it, so you can see which channels actually earn their budget. The honest answer is that no attribution model is fully accurate. Every one of them is wrong in a predictable direction, because real buying journeys cross devices, channels and weeks that your tracking never fully sees. What you can do well is choose a model whose blind spots you understand, pair it with a few simple experiments, and use the result to make better spending decisions instead of hunting for a single source of truth.
I get asked for “proper attribution” more than almost anything else, usually right after a business has spent money on four channels at once and cannot say which one worked. So let me be straight about what these numbers can and cannot show you. I would rather do that than hand you a dashboard that looks certain and quietly points you at the wrong budget.
What attribution is really measuring
Attribution does not measure reality. It measures whatever your tracking managed to record, run through a rule you chose in advance. That distinction matters more than any single number on the report. When someone sees your Instagram post on Monday, searches your name on Wednesday, clicks a Google ad on Friday and finally buys after a WhatsApp exchange on Sunday, no model captures all four moments cleanly. Each model keeps some of that story and throws the rest away.
So the useful question is never “which channel gets the credit”. It is “which parts of the journey is this model ignoring, and can I live with that”.
The main attribution models, and what each one hides
There are six models you will meet in almost every analytics tool. None is correct. Each is useful in a specific situation and misleading in others. The table below is the version I wish more dashboards showed next to the numbers.
| Model | What it credits | Where it misleads you |
|---|---|---|
| Last click | The final touch before the sale | Erases everything that created demand, and flatters bottom-of-funnel channels like brand search |
| First click | The very first touch | Overpays for awareness and ignores whatever actually closed the deal |
| Linear | Every touch equally | Pretends a passing impression matters as much as a decisive demo |
| Time decay | Recent touches more than early ones | The decay rate is a guess, and it quietly buries top-of-funnel work |
| Position based | First and last touch most, the middle less | The 40/20/40 split is arbitrary, not something your data proved |
| Data driven | Credit modelled from your real conversion paths | Needs high, steady conversion volume that most Qatari businesses simply do not have |
Notice the pattern. The simple models are honest about being rules of thumb. The data-driven model sounds scientific, and it can be excellent, but it quietly assumes you feed it thousands of conversions a month. Feed it forty and it is guessing with a straight face.
Why the Gulf makes attribution harder than the textbooks admit
Most attribution writing assumes a clean, online, card-on-file purchase. That is not how a lot of buying works here. A serious enquiry often moves to WhatsApp within the first message. Plenty of deals close on a phone call, a showroom visit, or cash on delivery, none of which your pixel ever sees. When the final, decisive step happens off the website, last-click quietly hands the credit to whatever brought them to the site and calls it a day.
Privacy changes made this worse for everyone. Between iOS limits on tracking, browsers dropping third-party cookies, and consent banners that many people decline, a meaningful share of journeys is now invisible to your analytics by default. If your reports look suspiciously confident about a WhatsApp-first, cash-friendly market, that confidence is the illusion, not the insight.
Multi-touch attribution, and when it is worth the effort
Multi-touch attribution tries to solve this by crediting several touches along the path rather than just one. In principle it is the right idea. In practice it asks for three things at once: clean tracking across every channel, enough conversion volume for the maths to mean something, and someone with the discipline to maintain it for months. Miss any one and you get a complicated report that is no more true than last-click, just harder to question.
For a business doing hundreds of conversions a month with a real analytics owner, multi-touch is worth building. For most small and mid-sized companies here, it is effort spent polishing a model while the underlying tracking is still leaking. Fix the leak first. The elaborate model can wait until it would actually change a decision.
What I recommend instead of chasing perfect attribution
I treat attribution as one input among several, not the referee. The combination that holds up is boring, and it works. Start with clean conversion tracking on one real revenue event, not eight vanity events. Add a simple “how did you hear about us” field at the point of enquiry, because self-reported source is unfashionable and quietly one of the most honest signals you have. Then, a few times a year, run a holdout: pause a channel in one region or period and watch what happens to revenue rather than to clicks.
Those experiments answer the only question that pays the bills, which is what happens to sales when I change spend. A model can suggest where to look. A holdout test tells you whether you were right, and when the two disagree you trust the test. Judge the whole thing on metrics your accountant recognises, revenue, cost per acquisition and margin, rather than on a prettier assisted-conversion chart.
The smallest first step that actually moves this forward
If your attribution feels like guesswork, do not start with a model. Start with one clean, trustworthy conversion event tied to money, wired correctly to every channel you run. That single fix removes more confusion than any multi-touch project, and it is usually the difference between numbers you argue about and numbers you act on. Slow is normal here. Invisible is not, and leaking tracking is how good marketing stays invisible.
Frequently asked questions
Which attribution model should I use?
Start with last-click because it is simple and everyone understands it, then put first-click and linear beside it to see how much demand creation you are hiding. The gap between them is the real story. Do not adopt data-driven attribution until you are past a few hundred conversions a month, or it will just dress up a guess in convincing clothes.
Is Google Analytics enough for attribution?
For most businesses here, yes, as a starting point, as long as your conversion tracking is clean and tied to real revenue events. GA4 lets you compare several models, which is genuinely useful. Its blind spot is the same as everyone’s: it cannot see the WhatsApp chats, phone calls and offline closes that finish a lot of Gulf deals.
Why do my channel numbers never add up to total sales?
Because journeys overlap and some touches are invisible, so every platform claims credit for buyers that other channels also touched. Add up each platform’s self-reported conversions and you will usually “sell” more than you actually did. That is normal, and it is exactly why you compare models and run holdout tests instead of trusting any single platform’s own count.
Does attribution still work with cookies going away?
Less cleanly than before, and that is not going to reverse. The shift is towards modelled and first-party data: your own CRM, logged-in behaviour, and honest self-reported source at the point of sale. Businesses that invested in owning their customer data are handling the change far better than those who leaned entirely on third-party pixels.
How much should I spend building attribution?
Far less than most vendors would like. The high-value work is clean tracking on one revenue event and a habit of running holdout tests, both of which are cheap. Elaborate multi-touch platforms only pay off at real volume with someone to run them, so buy that stage when you reach it, not before.
Turn your numbers into decisions you can trust
Good attribution will not hand you certainty, but it will stop you funding the wrong channel by accident, and that alone usually pays for the work. If you want conversions you can actually measure, my conversion rate optimisation process is the natural next step, and getting the budget split right across channels is easier once your tracking is honest. If you would rather have someone own this end to end, that is part of what a fractional CMO does. Book a free 30-minute call through the contact page and we will look at what your current numbers are really telling you, with no pressure either way.